PPO-based optimization of task offloading delay and battery cost

In recent years, the issue of task offloading in edge computing has attracted considerable attention, especially in the context of Mobile Edge Computing (MEC) servers. MEC servers offer substantial computational power and deployment flexibility, which are essential for minimizing task latency. However, the extended operation of these servers leads to accelerated battery degradation, thus increasing the associated battery usage costs. To mitigate this, offloading tasks to cloud servers or terminal devices has been identified as an effective strategy to reduce the MEC server's battery burden. However, this offloading process introduces additional latency, complicating the task of balancing offloading delay and battery usage costs. To address this challenge, this paper proposes a joint optimization framework that concurrently considers both task offloading latency and MEC battery usage costs. The framework quantifies battery degradation costs using a battery model. Given the non-convex optimization nature of the problem, directly solving it is complex. Consequently, the paper models the task offloading and battery cost optimization problem as a Markov Decision Process (MDP) and applies the Proximal Policy Optimization (PPO) algorithm from reinforcement learning to derive an optimal solution. This method allows for an effective trade-off between task offloading latency and battery usage costs. Experimental results validate that the proposed method significantly enhances performance, achieving an optimal balance between these two critical factors.